Welcome to Health Care POV | sign in | join
David Plaut: Off the Cuff

How Significant is ‘Significant'?

Published January 29, 2014 4:38 PM by David Plaut

Here are 12 hemoglobin values from each of two instruments including the means, minimum and maximum as well as the difference between the pairs and the % difference between the pairs:

   Instr. A Instr. B  A-B  %(A-B) 
 1  13.5  13.4 0.1   0.3
 2  11.0  10.9  0.2  1.0
 3  14.5  14.5  0.0  0.1
 4  15.9  15.8  0.1  0.5
5 9.7 9.7 0.0 -0.1
6 14.2 14.1 0.0 0.1
7 15.3 15.2 0.1 0.5
8 13.9 13.7 0.2 0.9
9 14.3 14.3 0.0 0.2
10 15.1 14.9 0.1 0.6
11 16.1 16.0 0.1 0.4
 12  15.0 14.8   0.2 0.8 
Mean 14.0 13.9 0.1 0.4
Min 9.7 9.7
Max 16.1 16.0
t-test = 0.000516

According to the t-test (paired) these two instruments are "highly significantly" different.  Although the average difference between them is about 0.1 gm/dL (0.4%).  Barely the CV of the methods.  Too often you will see data being significant in a statistical sense, when clinically it is not of clinical significance.  Since the editors and authors are not likely to change, we must be careful readeers. 

Oh, and by the way, using the t-test (unpaired) there is no difference between the two methods.  It seems that in some cases, the statistic chosen has a significant effect on the interpretation of the data.

posted by David Plaut

0 comments

leave a comment



To prevent comment spam, please type the code you see below into the code field before submitting your comment. If you cannot read the numbers in the image, reload the page to generate a new one.

Captcha
Enter the security code below:
 

Search

About this Blog

Keep Me Updated